2 Dispositions
2.1 Semantics
2.1.1 Core features of dispositions
A sugar cube is disposed to dissolve in water. We say that it is soluble or, slightly more contrived, that solubility belongs to its features. Likewise, we describe a match as being inflammable or that it has the disposition of inflammability. Further examples from various sciences include being supra-conductive and being magnetic from physics, being reactive and being acidic from chemistry, being fertile and being predatory from biology, being dictatorial and being reform-oriented from political science and being market-oriented and being competitive from economics. Let us point out and discuss arise on the level of the orinary, and they are of the same kind as those on the scientific level.
Trigger and manifestation
What those objects to which we correctly ascribe a disposition have in common is that they react or behave in certain ways when they are in certain conditions. Yet until the moment of being triggered an object's disposition might well be dormant and its manifestation non-occurent: as long as a sugar cube is not in contact with water its solubility is somewhat hidden; as long as a match is not struck or near fire its inflammability won't show itself. Thus, when we wish to explain what an object's disposition is we typically say how the disposed object would react if it were triggered in a certain way. This has three far reaching consequences; non-observability, modality and production, of which, to begin with, we will only introduce non-observability.
Classical and Logical Empiricism revisited
Now, remember the classical Empiricist claim that knowledge about the external world can only be gained by sense experience. Also recall the early twentieth-century Logical Empiricists' radicalisation of this epistemic dogma, namely to turn it into a semantic doctrine: the Verificationist criterion of sentence meaning, which says that a sentence expresses something meaningful if and only if we can spell out which phenomena we would have to observe with our five senses to call the sentence true or false. In fact, the meaning of a sentence, according to the Logical Empiricists, is verification by means of observation.
We can immediately see how sentences that mention dispositional predicates are problematic for Logical Empiricists: such predicatesrefer to features of the world that are-unlike colours, for example-not directly observable. You see that a crystalline powder is white but you do not see straight away whether it is soluble. And even when objectsreact as they are allegedly disposed to do, it is questionable whether we observe their disposition or merely the trigger and then the effect: you seea vase sit on a shelf, than you see it being hit and fall and finally you find it broken on the floor. All this is observable, yet do you see its fragility?
(1) Non-Observability (NON-OBS)
So, should we not, as a consequence of this non-observability, simply drop dispositional talk from allscientific discourse? After all, if it looks as if it is not amenable to Verificationism should it not simply be discarded as senseless? Yes this won't do in the case of dispositions and the reason why this is so is analogus to the one Hume had regarding causation: because it figures so prominently in both our ordinary and scientific talk, it is hardly possible to dismiss causation as meaningless. Rather, the Empiricists have to show how its meaning can be spelled out in words that refer to observable features only. Hume did offer such an analysis for causation so that his ultimate message wasn't 'there is no causation' but, rather, 'there is causation, yet it is a little different from what you think it is'.
Something perfectly analogous is true for modern Logical Empiricists and dispositional concepts. Dispositions also play essential roles in the conduct of scientific practise and, so, Empiricists had better find a way to translate dispositional talk into a language that can fulfil their Verificationist criterion of meaning so that it becomes acceptable. This is of the practical implemementations of the Empiricists' overall aim to 'give a rational reconstruction of the concepts of all fields of knowledge on the basis of concepts that refer to the immediately given'.
Reaching this goal might not seem a major problem in the case of dispositions. Take inflammability or solubility: we have clear intuitions how to phrase an 'if...then...'. sentence that specifies the observable conditions under which the entities that are soluble or inflammable show visible phenomena, namely to dissolve if put in water or to burst into flames if struck. These simple suggestions for a reconstruction/translation of sentences with dispositional predicates specify a method of verification: put it in water/strike it and see what happens. Does this dissolve/does it burst into flames? If yes then it is correct to ascribe solubility/inflammability to the object; if no then it is not.
Yet even if we assume that we can directly observe the respective stimuli conditions and reactions, our assumption that our goal is easily reached will prove to be wrong.
2.1.2 Void Satisfaction and modality
In order to see this, we turn to one of the classic papers in philosophy of science, namely Rudolph Carnap's Testability and Meaning. Together with Carnap, we:
- consider the question whether the so-called disposition-concepts can be defined, i.e. predicates which enunciate the disposition of a point or body for reacting in such and such a way to such and such conditions, e.g. 'visible', 'smelleable', 'tearable', 'soluble', 'indissoluble', 'etc. (carnap 1936: 440)
In accordance with the informal guesses we have made above, the most straightforward formal attempt of a definition of, for example, solubility is the following
simple conditional analysis ([SCA])
- [SCA]:
Suppose we wish to introduce the predicate 'D' meaning 'soluble in 'water' Suppose further, that 'T' and 'R' are already defined in such a way that 'Tx' means 'the body x is placed into water' and 'Rx' means 'the body x dissolves'. Then one might perhaps think that we could define 'soluble in 'water' in the following way: 'x is soluble in 'water' is to mean 'whenever x is put into water, x dissolves', in symbols: D(x) = def. T(x) → R(x) (Carnap 1936: 440)
The material implication
Carnap is perfectly aware from the outset that [SCA] is untenable and he later continues with his own, slightly exactly and why it fails, we need to have a brief look at the logical sentence connective →, i.e. the material implication it crucially employs. Readers who are unfamiliar with elementary formal logic should consult an introductory book before they continue reading. For everyone else, a reminder of the following characteristics of → will suffice: (1) The material implication belongs to the realm of extensional, truth-functional logic. (2) The truth conditions of the material implication are: p → q is true if and only if p and q are both true or p and q are both false or p is false and q is true. p → q is false if and only if p is true and q is true. p → q is false if and only if p is true and q is false. (3) A consequence of these truth conditions is that the material implication is confronted with the paradox that the falsity of p alone (no matter what's with q) as well as the truth of q alone (no matter what's with p) imply the truth of p → q.
Void Satisfaction (VC)
Returning to Carnap and the first tentative formal analysis of dispositional predicates he discusses - namely [SCA]: D(x) = def. (T(x) → R(x))^6 - we see that features (2), and, thus, (3) threaten the definition in the form of two famous phenomena: the so-called Void Satisfaction (VC) and the so-called ***Random Coincidence (RC) ***. We start with (VC) (and turn to (RC) later).Carnap knew that [SCA] will not do. In Testability and *Meaning, from which we have quoted already, he asks us to imagine a match, a, which has, as a matter of fact, never been in touch with water and never will have the chance to because, as it happens, we are about to burn it. THhen, as Carnap could easily show, the sentence T(x) → R(x) 8which defines for us what it is for something x to be soluble) comes out true for the match a. Thus, we would have to say that it is also true that the match is soluble, for this is what we defined solubility to be according to [SCA]. Yet we know that wooden things like match a are not soluble. An untenable result.Let's see why it is that T(x) → R(x) indeed comes out true for match a. First, we replace the dummy x with the specific a. Then T(a), the first part of the resulting sentence T(a) → R(a), has been, is and will always be false (we could write down ⌝ T(a)): the match has never been put in water and never will be because we have burned it. It has lost its chance to be ever exposed to water. Now, no matter whether the second part R(a) is true or not, the whole sentence T(a) → R(a) is already true according to how the material implication is defined. In other words, the match unwarrantedly satisfies the [SCA] criterion for being soluble. We speak here of a void satisfaction because the match fulfils the test for solubility without really having been tested. Some might be tempted to say that we simply have to bite the bullet and accept that the match is/was soluble, for who cares now that it has been burned? Yet one further thought reveals how devasting the (VC) really is: define the contradictory, opposite disposition being insoluble as T(x) → ⌝ R(x) and match a will voidly turn out to have been insoluble as well! Yet, it can hardly have been both soluble and insoluble.
Different conditionals instead of →
The following suggestion for a solution springs immediately to mind when we are confronted with this result: if the material implication, →, has such consequences it cannot be the correct formal representation for the 'if… then… ' sentence we have intuitively in mind when thinking of dispositions. In fact, the complaint continues, did we not say from the outset that we implicitly mean with attributions of dispositions is something like this: 'if match a were or had been put into water then it would (have) dissolve(d)'?
That is, we phrase the 'if… then… ' sentence that belongs to disposition ascriptions in so-called counterfactual or subjunctive terms. Unlike the material implication, such conditionals do not cling to the mere facts and (also) say something about possible and maybe not actual (i.e. counter to the facts) states of affairs. Transcending the mere de facto truths and falsities - here, the de facto absence of the putting in water of the match - counterfactuals take into consideration what would have been the case had we done it.
The latter is precisely what the material implication does and cannot do because of its truth-functionality: the mere fact of the absence of the test fixes the truth value of the material implication (to truth) regardless of what would have happened had we really done it. In this light, the (VC) problem turns out to have been foreseeable. The material implication, so it seems, cannot be successfully used to formalise what we mean by disposition ascriptions.
(2) Modality (MOD)
We can rephrase what has just been said to lead straight to the second feature - Modality (MOD) - of dispositions: objects that are disposed to react in certain ways when triggered seem to have some modal connection to possible (future) manifestations. So-called modal statements typically say what is possible or what is necessary, or, as in the case of dispositions, what would be the case if something else were to happen: 'Instead of this book you could read a novel' is an example of possibility; 'necessarily, everything that is red has a colour' an example of necessity; and 'if you were to put this into water it would dissolve' for the counterfactual modality. By contrast, non-modal statements (sentences with the material implication, for example) only trace what is de facto the case (or not the case) without any claim about what is possible, necessary or would happen under certain circumstances.
As a brief interlude, consider three further facts about modality. We will get back to many of those features in due course.
First, necessity and possibility are interdefinable: if something p is necessarily the case (for example, that red is a colour) then it is not possible for p not to be the case. Or, the other way round, if it is possible that p (for example, that you hold this book in your hands) then it is not necessarily not the case that p. So, if we know everything about necessity we also know everything about possibility. Second, there are interesting relations between modal and non-modal statements. For example, if it is the case that this book has fewer than 500 pages then it is also possible that the book has fewer than 55 pages, for were it impossible for the book to have fewer than 500 pages it would not have fewer. We'll leave these relations in the backgrund for now. Third, there are many kinds of necessity (and, thus, possibility) we could distinguish: logical necessity (the light is switched on or it is not switched on), conceptual or analytic necessity (all sisters are female), mathematical necessity (there is no largest prime number), metaphysical necessity (no object's surface is both red and green all over), moral necessity (one ought not torture anyone) and natural necessity. The latter we can describe intuitively (without taking this as a proper definition) as necessity due to how the world works. So, for example, if Einstein is right and no massive object can accelerate beyond the speed of light then it is naturally necessary that nothing moves faster than light.
Coming back to our core issue of dispositions, note that (NON-OBS)-non-observability, the first of our three features of dispositional - and (MOD) are not unrelated. On the contrary, the (NON-OBS) of dispositions is (partly) a consequence of their (MOD): you can observe with your own eyes only what is factual, what is here and now in front of you. Modal facts, such as what would happen if something else had happened, transcend the directly-observable. Simon Blackburn expresses this fact concerning modal necessity thus:
- Observations only extend to limited periods of space and time: how could we have within our view something that essentially casts its net over the whole of space and time?
Similarly, we do not observe what disposed objects can do or must do or what connection they might have to unactualised events or states of affairs.
A different formulation from [SCA] but retaining →
Carnap and the logical Empiricists were, of course, perfectly aware of all this, i.e. of the fact that dispositionality carries some modal force, that dispositional modality is hardly observable with your bare eyes (i.e. that, here, (MOD) leads to (NON-OBS)) and that, as a result, dispositional ascriptions are not so easily paraphrased using observational terms and the material implication as interpretation of the implicit 'if… then…' statement.
Yet adherence to their Verificationist doctrine meant the Empiricists had the strong urge to continue to try to spell out dispositionality in observational terms and, also, by sentence connectivities whose truth conditions do not lead us onto non-truth-fictional territory. Triumph in these endeavours would be welcome for the following reason: if it is straightforwardly observable whether T(x) and R(x) are true or false then the truth of T(x) → R(x) (or some slightly more complicated version of it) can be easily calculated: only if T(x) is observed to be true but R(x) observed to be false is the combined sentence false, and true in all other cases. Thus, if some (more complex) definition with a material implication turned out to be acceptable after all the whole exercise would be a paradigm case of successful Logical Empiricism at work, i.e. prove empirically (by observation) which truth value the simple sentences have, then calculate logically what truth value the complex sentence has. So, the extensional, truth-functional definition of → suits Logical Empiricism very well indeed.
Thus, Carnap and the other Logical Empiricists reacted to the (VC) by keeping the material implication after all, yet reformulating the simple T(x) → R(x) sentence into a slightly more complicated definition than [SCA] is.
Finally, compare the observability and modality issue regarding dispositions to Hume's view on causation (Section 1.2): Hume fought the Rationalists' conception of causation, which said the causal relation is a necessitating one - where, if the cause c occurs so does, with necessity, the effect e - and offered a definition of causation that was in no need of reference to modal necessitating forces. In the light of this we can see the Logical Empiricists' further endeavour to define dispositional predicates, come what may, with the material implication as an attempt not only to bypass the (VC) but also as the hope to give an analysis of dispositions without the need to refer to modal counterfactual facts.